config.json

{
    "seed": 1234,
    "epochs": 10000,
    "learning_rate": 0.0002,
    "betas": [
        0.8,
        0.99
    ],
    "eps": 1e-09,
    "batch_size": 32,
    "fp16_run": true,
    "lr_decay": 0.999875,
    "segment_size": 8192,
    "init_lr_ratio": 1.0,
    "warmup_epochs": 0,
    "c_mel": 45,
    "c_kl": 1.0,
    "grad_clip": null,
    "min_seq_length": null,
    "max_seq_length": 400,
    "min_spec_length": null,
    "max_spec_length": null,
    "min_speaker_utterances": null,
    "last_epoch": 1,
    "global_step": 1,
    "best_loss": null,
    "audio": {
        "filter_length": 1024,
        "hop_length": 256,
        "win_length": 1024,
        "mel_channels": 80,
        "sample_rate": 22050,
        "sample_bytes": 2,
        "channels": 1,
        "mel_fmin": 0,
        "mel_fmax": null,
        "ref_level_db": 20,
        "spec_gain": 1,
        "signal_norm": true,
        "min_level_db": -100,
        "max_norm": 1,
        "clip_norm": true,
        "symmetric_norm": true,
        "do_dynamic_range_compression": true,
        "convert_db_to_amp": true,
        "do_trim_silence": false,
        "trim_silence_db": 40,
        "trim_margin_sec": 0.01,
        "trim_keep_sec": 0.25,
        "scale_mels": false
    },
    "model": {
        "num_symbols": 47,
        "n_speakers": 18,
        "inter_channels": 192,
        "hidden_channels": 192,
        "filter_channels": 768,
        "n_heads": 2,
        "n_layers": 6,
        "kernel_size": 3,
        "p_dropout": 0.1,
        "resblock": "2",
        "resblock_kernel_sizes": [
            3,
            5,
            7
        ],
        "resblock_dilation_sizes": [
            [
                1,
                2
            ],
            [
                2,
                6
            ],
            [
                3,
                12
            ]
        ],
        "upsample_rates": [
            8,
            8,
            4
        ],
        "upsample_initial_channel": 256,
        "upsample_kernel_sizes": [
            16,
            16,
            8
        ],
        "n_layers_q": 3,
        "use_spectral_norm": false,
        "gin_channels": 512,
        "use_sdp": true
    },
    "phonemes": {
        "phoneme_separator": "_",
        "word_separator": "#",
        "phoneme_to_id": null,
        "pad": "_",
        "bos": "^",
        "eos": "$",
        "blank": "_",
        "blank_word": "#",
        "blank_between": "tokens_and_words",
        "blank_at_start": true,
        "blank_at_end": true,
        "simple_punctuation": true,
        "punctuation_map": null,
        "separate": [
            "\u02c8",
            "\u02cc"
        ],
        "separate_graphemes": false,
        "separate_tones": false,
        "tone_before": false,
        "phoneme_map": null,
        "auto_bos_eos": true,
        "minor_break": "\u00b7",
        "major_break": null,
        "break_phonemes_into_graphemes": false
    },
    "text_aligner": {
        "aligner": "kaldi_align",
        "casing": "lower"
    },
    "text_language": "en_US",
    "phonemizer": "gruut",
    "datasets": [
        {
            "name": "cmu_arctic",
            "metadata_format": "text",
            "multispeaker": true,
            "text_language": null,
            "audio_dir": "/media/12tb/en-us/cmu_arctic",
            "cache_dir": "/media/cache/cmu_arctic"
        }
    ],
    "inference": {
        "length_scale": 1.0,
        "noise_scale": 0.333,
        "noise_w": 0.333,
        "auto_append_text": "."
    },
    "version": 1,
    "git_commit": ""
}